LoopMoEVR restores ultra-HD videos across multiple degradation types

LoopMoEVR: Loop-Based Degradation-Aware Mixture-of-Experts for Unified UHD Video Restoration

Computer Vision and Pattern Recognition

Summary

High-definition videos can look bad because of things like haze, rain, noise, or low light. The authors designed LoopMoEVR, a new kind of video restoration model that can fix many of these problems using a clever loop-based approach. This model uses a small number of settings to understand how the video is degraded and then improves the video quality step-by-step. It works well on ultra-high-definition videos and outperforms other current methods despite using fewer parameters.

What this means in practice

  • For video streaming platforms: Enhance the visual quality of UHD videos suffering from haze, rain, noise, or low-light conditions during transmission or playback.
  • For film restoration studios: Restore damaged or degraded ultra-high-definition video footage with a unified tool handling multiple degradation types efficiently.

Authors

Yucheng Xin, Runci Bai, Yongcong Wang, Guangwei Gao, Jiao Liu, Dianjie Lu, Linwei Fan, Zhuoran Zheng

Abstract

Recently, unified high-definition image restoration has attracted considerable attention; however, existing models tend to excessively increase their depth in pursuit of improved generalization, which often yields only limited gains. Meanwhile, loop-based learning paradigms have drawn widespread attention due to their low parameter counts and strong regression capability, as exemplified by GPT-6 and looped Transformers. In this paper, we introduce the loop learning paradigm to address restoration tasks that require cross-domain learning. Specifically, we propose LoopMoEVR, a loop-based mixture-of-experts model capable of handling degraded ultra-high-definition (UHD) inputs. First, a degradation-conditioned low-rank loop embedding is designed to construct input-dependent stage conditions. Second, a spatio-temporal iterative adaptive normalization module, termed IterAda3DN, is developed to fuse local features with global loop context, thereby performing position-wise affine modulation. Finally, the expert branches further integrate the attention-updated local and global video states with the loop conditions to generate dedicated modulation parameters, while an input-conditioned depth predictor adaptively configures the number of loop iterations. With only approximately 0.884M trainable parameters, the proposed model uniformly handles UHD video dehazing, deraining, denoising, and low-light enhancement tasks, achieving state-of-the-art restoration performance on both public benchmarks and real-world scenarios.